Sunteți pe pagina 1din 28
The Assembling of "Stria" by John Chowning: A Philological Investigation Author(s): Laura Zattra Source: Computer

The Assembling of "Stria" by John Chowning: A Philological Investigation Author(s): Laura Zattra Source: Computer Music Journal, Vol. 31, No. 3, The Reconstruction of "Stria" (Fall, 2007), pp. 38-64 Published by: The MIT Press Stable URL: Accessed: 06-04-2016 05:28 UTC

REFERENCES Linked references are available on JSTOR for this article:

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact

information about JSTOR, please contact The MIT Press is collaborating with JSTOR to digitize,

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to Computer Music


This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Laura Zattra

Dipartimento di Storia delle Arti Visive e della


Universita di Padova

Piazza Capitaniato 7

Padova, Italy 35139

This article starts from the assumption that musi- cology needs methods borrowed from philology for studying computer music. The analysis of the cre- ative and revision process that John Chowning car- ried out in the realization of Stria (1977) is made possible by textual criticism and interpretation based on digital and audio sources, sketches, and

oral communications.

This research moves from the hypothesis that

Stria exists in several versions and follows the his-

torical genesis of the assembling of the sections. Conclusions trace the various stages of the assem-

bling process, from the very first synthesis of the

piece to two different and coexisting four-channel and stereophonic versions.

Theoretical Introduction

The electroacoustic music work is an " apparently unclassifiable object," fragmented between several human and technological agents, different moments of creation, realization, and performance, and vari- ous theoretical concepts (Barriere 1990). Any musi- cological study must start from the heterogeneous documentation of the compositional process, which

includes compositional sketches and various types

of scores - if one considers the concept of "text" as

a physical object where the sign represents the


Philology of music states, in fact, that "text" is a "reference model" - the place where projects are

written in a tentatively stable form to preserve and

transmit it (Caraci Vela 2005). Nevertheless, the

text within electroacoustic music is not necessarily

a visible trace. As Angela Ida De Benedictis writes

about analog music, text is also the support - the tape - that preserves the sound (De Benedictis 2004). In computer music, it is also the data-storage device and digital data with which a machine gener-

Computer Music Journal, 31:3, pp. 38-64, Fall 2007

© 2007 Massachusetts Institute of Technology.

The Assembling of Stria by

John Chowning: A

Philological Investigation

ates a computation that is not an immediately intel-

ligible symbol. Moreover, electroacoustic music

often implies the role of the musical assistant,

or the performance assistant, or someone who trans-

lates compositional ideas into digital data: These

all are human agents who carry on a codification

through the processes of writing. The study of the

process of composition (the term ecriture would be

more appropriate) and the text must be applied to

the tape, the CD, digital memory, the digital score,

and so on, because "these texts do not certainly

show a visible context of writing, but do certainly

maintain its essence and its reproduction technol-

ogy" (De Benedictis 2004, p. 247).

As the last resort, among these different texts

converging to the electroacoustic work, and within this actual culture where aurality-orality and ecrit- ure are blurred, the composer's "mental" text is fundamental. It represents the musical intention,

and it is worth being analyzed if one considers it as

a support for the writing of the musical thought.

Obsolescence and preservation are crucial problems in the study of electroacoustic music. Therefore, mental texts (of composers, technicians, etc.) are important to the preservation and analysis of musi-

cal works.

Analysis of the sources is therefore the primary

method to study electroacoustic music: It is a ques-

tion of examining the texts as a process of writing,

conceived to reach the electroacoustic "arborescent

reality" (as defined by Barriere 1990). This is the re-

frain of my investigation on Stria by John Chown-

ing, "a music which did not resemble any other

music, touching for the beauty of its relaxation and

sound concatenation" (Risset 2005, p. 50). Together

with the presence of objective witnesses, the anal-

ysis benefits from and is forced to take into account

the presence of the author, who can help elucidate and sometimes complicate the analysis with his memories. These types of sources of course need to

be interpreted, verified, and compared with physical

sources, but are fundamental to allowing the anal-

ysis to proceed. Without the presence of John

38 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Chowning, his comments, recollections, feedback, and supervision, this research could often have en-

countered a "dead end." This approach occasionally

obliges the musicologist to "mix" philological and

historical analyses to reconstruct the compositional

process. Even if that could seem a confusion of plans and a renunciation of adopting an objective consideration of the sources (particularly audio sources), on the other hand it is necessary. However,

this is why I do not go into details - it goes beyond

my competence - concerning physical aspects of support material (acetate or other), condition (state

of degradation of the tape, etc.), state of preserva-

tion, number of segments of the tape, and so on. I

consider only elements and aspects useful for my purposes, which are the development of the assem- bling process of the musical piece.

Stria: A Short Overview

In 2005, the GRM (the French Institut National de l'Audiovisuel) published the volume John Chown-

ing: Portraits polychromes (Castanet et al. 2005). The book presents a collection of interviews with the composer, as well as articles on his musical

works and the impact of his research. Moreover, it

contains some musicological analysis of two of the

four pieces (Turenas and Stria) that Mr. Chowning

composed between 1971 (Sabelithe) and 1981

[Phone). Bruno Bossis's essay in this volume draws on a

previous article by Matteo Meneghini and recon-

structs various aspects of the realization of the piece in terms of macrostructure, frequency sys-

tem, Golden Mean, and so forth (Meneghini 2003;

Bossis 2005). Nevertheless, because the philologist

pays attention to different sources, some aspects re-

lated to the text of Stria emerged that make the piece

worth being further analyzed. After Meneghini's ar-

ticle and Borghesan's attempt to re-synthesize the

piece following Borghesan (2005), I discovered some additional aspects of the piece, and new details and discrepancies emerge in the versions. As a matter of

fact, analysis reveals at least three versions of Stria, which shows this is a very interesting case from a

philological point of view.

Stria is the result of the musical application of Mr. Chowning's research into frequency modulation (FM). October 2007 marks the 30th anniversary of the completion of Stria, and it also marks the 40th anniversary "since I stumbled on the FM synthesis algorithm!" (Chowning 2006). Mr. Chowning re- cently reminisced about the precise circumstances and historical dates during a seminar and concert he gave in Buenos Aires together with Jean-Claude Ris-

set and Max Mathews. On that occasion, Mr. Risset

showed to the audience some notes dated 18 Decem-

ber 1967 that he made when Mr. Chowning visited

Bell Laboratories and described the FM data from

some personal undated notes. Mr. Risset's notes are useful to establishing Mr. Chowning's sketches and

their discovery: the origin of the FM applied to

sound synthesis dates Fall 1967. (As an historical detail, Mr. Chowning adds that he had visited Bell Labs also in Summer 1967, but he has no recollec- tion of mentioning FM to Jean-Claude Risset and

Max Mathews, which he surely would have done

had the discovery already been made.) Mr. Risset later used FM synthesis in part of his piece Muta- tions, which was completed in 1969 and involved Mr. Chowning's work with the moving sound- sources project. Mutations, then, is the first compo-

sition in which FM was used (Chowning 2006).

In 1974, after a sabbatical year during which he wrote his famous article on FM synthesis (Chown-

ing 1973), and after being informed by Stanford Uni-

versity that his position was not renewed (Means

2005; Castanet et al. 2005), Chowning came to

Berlin with a DAAD fellowship for a research stay of one year (as composer in residence) at the Ger- man Arts Academy. Mr. Chowning obtained this position through the recommendation of Gyorgy Ligeti, who was aware of his situation at Stanford. At the Technical University, he could use only a PDP-10 computer with no sound converter, so he decided to study deterministic processes (Roads 1985). He worked deeply on the FM technique ap- plied to composition, especially discovering that a carrier-to-modulator frequency ratio composed of powers of the Golden Section yielded low-order side-band components that are also powers of the

Golden Section.

Mr. Chowning thought about the composition



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

with such intensity that when he finally decided to realize his musical project, he completed the work

in only four months at the Stanford Artificial Intel-

ligence Laboratories, where he had returned. This

occurred during the summer of 1977, from July to

October, thanks to an IRCAM commission. In fact,

this was one of IRCAM's first composition commis- sions, as the musical production in Paris had started the year before (Zattra 2003). It was also one of the

very first computer-music pieces for solo tape com-

posed at IRCAM. (Also that year, Jean-Claude Ris- set realized Irihaimonique, for voice and

synthesized sounds (Lorrain 1980).) Stria was presented on 13 October 1977 at 8:30 PM, at the Centre Pompidou (Grande Salle Polyva-

lente). The concert series also included Luciano Be-

rio's Sequenza for voice and electronic, together with premieres by H. Pousseur, M. Decoust, and J. Druckman. The concert was part of a presentation by Berio entitled "La Voix des voies," and within

the Exposition electroacoustique by Luciano Berio

(14 events conceived as "spectacle audiovisuel"

with commissions from Decoust, Denisov, Druck-

man, Eloy, Pousseur, and Stockhausen) (Riviere and

Pouillon 1976). The Passage du XXe siecle cata-

logue indicates another concert on 1 October (con-

cert "Exposition electroacoustique"), with music by

Mr. Chowning, Mr. Stockhausen, Mr. Druckman,

and Mr. Pousseur (Riviere and Pouillon 1976). Nev-

ertheless, Mr. Chowning notes that he arrived in

Paris only a day or two before the 13th and stayed

with Jean-Claude Risset and his family (Chowning


Stria corresponds to an intermediate point be- tween parameter-by-parameter composition and au-

tomatic composition. Mr. Chowning declares that

during the composition of the piece, he seemed to always want more and more control over detail and

ended up for this reason with a large number of vari-

ables. Because he had spent many hours thinking carefully about controlling a larger-scale formal structure, he then wrote an algorithmic procedure called event 2 that brought an even greater reward in compositional inspiration and control. Mr. Chowning believes that computers are fundamen- tally capable of providing more than is requested or envisioned, so that one feels, as a composer, enor-

mously empowered - "not surprising, when one

considers that a given hardware and software repre- sents tens of thousands of human-years of thought

about thought and invention" (Chowning 2004).

Automatism and the extreme control of each mu-

sical parameter related to the Golden Mean could lead one to think of a serial approach to Chowning's music. He concedes that this system is interesting, but all the radicalism of that avant-garde approach did not completely satisfy him. He rather looks at the organization techniques of the musical material

of 1,000 years of Western music history: spatial or-

ganization of pitches, counterpoint, and control of the harmonic and inharmonic spectra. Neverthe- less, all those dimensions must always be related to perception, which remains the central moment of the musical verification (Gayou 2005).

The Golden Mean is the reference ratio that con-

trols the spectral (vertical) space (Meneghini 2003; Bossis 2005), and the related Fibonacci sequence in- spires the overall timing of the piece (horizontal space); the two are closely related. Stria is "a com- pletely abstract construction" and non-referential. The composer writes that it is "something that could be done by a computer but could not be done

by any other electronic device. With any sort of

analog synthesizer, I don't think it's possible just be-

cause they don't have the precision or the program-

ming as part of their structure. It's a piece that is

most uniquely tied to the digital domain" (Chown- ing 2005c).

General History of the Assembling Process

At the beginning of its compositional process, Stria was conceived as a piece that should have been real-

ized in real time. It was originally planned for the

Samson Box, a real-time digital synthesizer designed by Peter Samson of Systems Concepts Company. Peter Samson began his research in 1974 to develop

a large-scale digital sound synthesizer controlled by

a mainframe computer, capable of producing 256 in- dependent sound sources in real time (Roads 1996, p. 918). The Samson Box was not finished in time,

however (it was delivered to the Stanford Center for

Computer Research in Music and Acoustics -

40 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

CCRMA- in late 1977). Thus, the piece was real-

According to the Computer Music Journal 1:3


in sections, converted and recorded from the

(June 1977), the first performance of the CCRMA

computer (a Digital Equipment Corporation PDP-

version was scheduled for 26-30 October 1977,

10) onto four-channel analog tapes - one tape for

where it was performed during the International


section - using a Scully four-channel half-inch

Computer Music Conference at the University of


recorder, in September and October 1977. Af-

California, San Diego (Chowning and Keislar 2007).

terward, Stria was never realized on the Samson

Box or any other real-time device. Today, a real-time

rending would of course be possible. Computation

was a problem when Stria was conceived, but now,

"78 oscillators (26 Stria instruments using 3 oscilla-


per instrument) would be easily achieved on


laptop, and I hope that a real-time version will

soon see the light" (Chowning 2005c). The composer could not realize the work onto a single tape at Stanford because, on the one hand, the PDP-10 did not have enough digital memory to store more than a certain number of samples, and on the other hand because CCRMA had only one

four-channel analog tape recorder. Mr. Chowning then took the tapes to IRCAM in Paris, where, as he

remembers, there were many four-channel recorders:

"It seemed to me then, they were used as door stops

even!" (Chowning 2004). At IRCAM, he was able to

mix the sections into the complete piece. No IRCAM

technician was involved in the production, except

for Andy Moorer, who had worked at Stanford and

at the time worked at IRCAM. Mr. Moorer, who had

given Chowning instruction in the SAIL (Stanford Artificial Intelligence Language) language four

months earlier, simply helped in the starting and stopping of tape recorders to make the final tape

(Chowning 2004, 2007).

Mr. Chowning remembers that he was not per-

fectly content with the timing of the mix of this first

version (Chowning 2005c). Besides, immediately

following the first performance, Luciano Berio and

Cathy Berberian went from the audience to speak to him. They seemed to like the composition, but Be- rio suggested - and Mr. Chowning agreed - that it needed little adjustments, particularly in the dura-

tion, which could be shorter in some points without spoiling its beauty and proportions. The morning following the concert (14 October), Mr. Chowning

reassembled the sections and truncated some other

to what he now calls the "CCRMA version," which

is the version that he returned with to Stanford.

In 1984, a performance of Stria took place at IR-

CAM, but I did not find any article or recording doc-

umenting the event. The IRCAM Web site records

two dates: 13 January and 21 June, both at the Es-

pace Libre (IRCAM 2005). There are no firsthand

accounts whether the version played was the first long version or the reduced version. However, Mr. Chowning believes that, as he remembers, there

was no other version at that moment than the re-

duced version (Chowning 2006).

In 1988, the Wergo label initiated the Digital Mu- sic Digital series on the recommendation of Johannes

Goebel. Wergo dedicated a CD to Mr. Chowning's

works (Chowning 1988). This was one of the first

series of digital discs in the world, conceived for the

preservation of early computer music. The project had started in 1986 involving Johannes Goebel, Max

Mathews, Patte Wood, CCRMA, and- from 1990-

ZKM, and it was named IDEAMA. Its goal was to

preserve the most important and endangered early electroacoustic works and make them publicly available. Stria was published together with Ture- nas, Phone, and Sabelithe. Through the course of my initial research, it had

been evident that Stria was realized in different

steps and has had different durations and descrip-

tions since its first conception. This soon meant

that I was facing an interesting musicological case

of a computer piece with multiform identity that

required philological investigation. The logical

aim of this study was therefore to try to answer the

following questions: In how many versions does

Stria exist? Are those versions different and author-

ized redactions of the piece, or did they escape Mr. Chowning's control? Which is the final version?


The sources shown in Table 1 represent the signifi- cant evidence and extant documentation (in some



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Table 1. Sources for the Study of Stria

Written sources

TbMl Description: Toby Mountain's graphical score

of the piece, unpublished

Reference: Mountain 1981

Duration: 15 '46"

TbM2 Description: Toby Mountain's sketches for the


Reference: Mountain 1980

TM-LK Description: short text by Tod Machover quot-

ing an unpublished interview with the com- poser by Lev Koblyakov

Reference: Machover 1984

Duration: No duration marked

Df Description: Dodge and Jerse's scheme; graph of

the compositional structure Reference: Dodge and Jerse 1985, p. 126

Duration: 18'

RD Description: Roberto Doati's analysis,- musico-

logical analysis of the compositional process,


Reference: Doati 1988

Duration: two durations emerge: 15 '46" [4-ch) and 16'57" (Wergo)

IR Description: IRCAM Web site (written docu-

mentation of the piece)

Reference: IRCAM 2005

Duration: 15'

Audio sources

4-ch Description: Roberto Doati's four-channel ver-

sion; digitization made from an analog four-

channel tape (Mr. Doati's personal archive)

Reference: Doati 1988

Duration: 15 '46"

WER Description: Wergo CD

Reference: Chowning 1988

Duration: 16'56"

IR-ta Description: IRCAM's four-channel tape

No dating on the box

Duration: unknown

IRdig Description: IRCAM's digitization from the

four-channel tape (CD-ROM provided by John Chowning, dated 6 January 2005); sampling rate

is 48 kHz

Reference: Mr. Chowning's personal archive

Duration: 17'26"

Table 1. (continued)

Computational sources

fCdig Description: handwritten notes, computer file

printout, and unpublished manuscript with dig-

ital data for the calculation of the structure of

the piece. Reference: Chowning 1977a-e, 1977(?)a,

1977(?)b, 1978a-b

Duration: each section (from TO .MEM to

END . mem) shows a different duration

cases not published) dedicated to the piece consid-

ered for this research. Recall that I state that an au-

dio support is also a text, therefore I consider here

audio texts in the same way as written sources.

They are all important for the study of the realiza-

tion process and its development. In most of the cases, audio sources are fundamental for the analysis

of the assembling because they demonstrate the var-

ious versions. Because my philological investigation

attempts to understand the assembling process, for each source I cite its duration when indicated, which will be the most important dimension for the development of the analysis. Of course, the audio physical source itself does not reveal much more

than its physical evidence. For its signal analysis

(e.g., its duration), it is necessary to pass through ex- ternal software for the editing of music: In this case

I use Cool Edit Pro software, version 1.2. The wit-

nesses are divided into written and audio texts,- for

each part, the material is organized in chronological

order according to the date of publication or diffu- sion of the source, not to its creation. The first part

lists the witnesses (textual analyses, articles, graph- ical analyses) in the literature completely dedicated

to the piece or, if part of a book, particularly signifi-

cant for its dissemination. The second column lists

audio sources and printouts of the computational

data. All witnesses are authorized sources that are

produced by Mr. Chowning himself or accepted and

partly controlled by him.

I compare at the same time audio and written documentation, because I am interested in the pure

timing data to reconstruct the history of the " com-

positional" process of the piece - intended as as- sembling of sections rather than synthesis. Five

42 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

possible durations of the piece are shown: The longest duration is 18 minutes (D/); the second- longest is 17'26" (IRdig)} the third duration, 16'56", corresponds to the CD version by Wergo [WER, also cited in RD)} the fourth lasts 15'46" (TbMl and 4-ch)} the last is 15 minutes (IR). The presence of two du- rations in the same article (RD; see Doati 1988) is

an indication that at least two versions of Stria have

co-existed. Some of these sources, such as TM-LK,

are worth being presented even if not directly useful to the analysis of different durations, because they help by adding information to the historical circum-

stances. IR-ta is still a problematic source.

Other sources - some of them unpublished -

partly or completely dedicated to Stria include

Baudouin (2006), Borghesan (2005), Bossis (2005),

Boulez (1980) (audio recording), Chowning (1977b),

Chowning (2005a), Chowning (2005b), Dahan

(2001), Gayou (2005), Means (2005), Meneghini

(2003), Risset (1988), Risset (2001), Risset (2005), Roads (1985), and Schoeller (1986), as well as Jo- hannes Goebel's liner notes accompanying the Wergo recording (Chowning 1988). These have been

useful from time to time for their historical, aes-

thetic, technical, or analytical aspects, but not di- rectly used in my philological research on the assembling of the piece.

Description of the Sources

Toby Mountain's Graphical Score (TbMl)

I did have knowledge of Toby Mountain's notational representation from an unpublished musicological analysis made by the Italian composer Roberto

Doati, in 1988 (Doati 1988). The analysis ends with

Mr. Mountain's score, which considers a duration of

15 '46". An introductory page not shown here ex-

plains that the score was produced solely from aural

observation without the use of program data. In this

sense, it is intended to accompany the tape and to

be a general guide for the listener (underlined in the

text, TbMl). Figure 1 shows the graphical score. In 1980, Toby Mountain was a graduate student in music composition at the University of California, Berkeley, and received his PhD in 1981. The Stria

analysis was part of his thesis. Mr. Mountain's anal-

ysis tends to translate the pitch information of the

electronic piece into traditional notation.

Toby Mountain's Sketches for the Analysis (TbM2)

Toby Mountain kindly gave me 1 7 digital copies of sketches he did for the analysis of the piece (Moun-

tain 1980). Four of them include the original graphi-

cal score of Figure 1 . All of these sketches are

undated. This fact sometimes poses analytical prob- lems, because the details regarding the sections of the pieces diverge. Nevertheless, they are very use- ful because they show some aspects that are lacking

in other sources.

Tod Machover and Lev Koblyakov (TM-LK)

In August 1984, Tod Machover wrote a short text

for IRCAM's internal documentation of research in

which he describes Mr. Chowning's work. The doc- ument is important because it specifies the research period, as it was formally documented at the French center ("Date of Project: Proj.l: 1976; Proj.2: 10/78

to 7179" , where the first period is for Stria and the

second for Phone). It also says that Stria is one of IRCAM's first commissions, even if the work itself was produced at Stanford and not at IRCAM. Mr. Machover complains about the lack of any serious analysis dedicated to the piece at that time "and Lev Koblyakov's short article can only serve as a

basic introduction" (Machover 1984, p. 17).

Mr. Chowning himself contacted Lev Koblyakov to learn where to find the analysis to which Mr. Ma- chover refers. Mr. Koblyakov remembers a meeting with Mr. Chowning in Spring 1979, when he was invited by Pierre Boulez to give a series of lectures on the evolution of music in the 20th century

(Chowning 2005c). Mr. Chowning showed at that

time some "programs for Stria," but he did not give them to Mr. Koblyakov. Afterward, in 1984, Tod Machover invited Lev Koblyakov to write an essay on the piece, but the article was never published. He unfortunately cannot recover the article since he moved from Jerusalem (Koblyakov 2005).



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 1. Toby Mountain's notational representation of Stria (Mountain 1981).

notational representation of Stria (Mountain 1981). Another important note in Tod Machover's ar- ticle says

Another important note in Tod Machover's ar- ticle says that no score exists and that most of the

composer's data is no longer available. He also adds

that "one is left only with the possibility of doing

an aural analysis of the piece, which is virtually im- possible at any but the most superficial level, since

the piece deals entirely with gently shifting inhar-

monic spectra generated by frequency modulation" (p. 17). Even if Mr. Machover's text does not show any description of the piece and its duration, this source is important, because it states that he did not have the ability, in 1984, to look up the computer data. This means that original digital data was re-

moved from IRC AM (or got lost after the first per-

formance), where John Chowning remembers he left a copy of everything (Chowning 2006).

Dodge and Jerse's Scheme (DJ)

The famous book Computer Music published in

1985 by Charles Dodge and Thomas Jerse discusses the composition as well. The passage related to

Stria very briefly explains the use of the Golden

Mean, the frequency space, and the overall shape of the piece. The well-known scheme illustrated in

44 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 2. The scale compo-

sitional structure of Stria

(Dodge and ferse 1985, p. 126).

Figure 2 shows different blocks, one of which is

shown in the insert to the figure, representing a

number of tones that enter in temporal Golden Mean proportions. Dodge and Jerse consider this figure a sketch of the shape (without any claim to be precise in details) of the 18-minute piece. They describe the general sonority with predominantly long sounds, with an almost complete absence of

percussiveness. Even though there are some

changes in location of the sound, the movement

is very gradual.

Roberto Doati's Analysis (RD)

In 1988, Italian composer Roberto Doati analyzed

Stria. This was the first analysis of the piece (un- fortunately never publicly diffused) focused on technical aspects and on original handwritten docu- mentation by John Chowning (use of the Golden Section, FM instruments, and spectral space divi-

sions). At the beginning of his text, Mr. Doati re- marks that the duration of the four-channel tape

he possesses, given by Mr. Chowning himself, is

15 '46", while the CD version (he refers to the

Wergo CD), has a 16'57" duration (Doati 1988). He

does not give any explanation of this discrepancy. The paper develops a technical analysis of the spec- tral space and the general use of the Golden Mean in the piece. The last part presents an analysis of the listening with segmentation using some audi- tory images. In the last pages, Mr. Doati shows Toby Mountain's graphical score to support his own listening.

IRCAM's Web Site [IR)

The IRCAM Web page devoted to Stria indicates a

15' duration. The page briefly introduces the piece

and reveals some difficulties in the demonstration

of the compositional aspects by means of traditional analytical tools. The unknown author of the article

regrets the " impossibility" of analyzing the pres-

ence of the Fibonacci series. Moreover, the author finds it impossible to trace the Golden Mean in the

overall spectrum, even employing a computer pro-

gram to analyze two different sequences of the first

sequences of the piece. Nevertheless, the IRCAM

analysis succeeds in finding some regularity in the beginning of some adjoining spectra and in their du- rations. (For example, after 0'55", there is a clear changing of spectrum.) The macro-proportions are

difficult to find: It is possible, however, to identify

the main powerful climax in the middle of the piece (from 8'00" to 9'00"). Although it does not mention the version analyzed, this element of the article would relate it to the four-channel version (as one could easily see later in Figure 14). According to the analysis, the overall form seems to be an arch. This

short article does not make use of technical infor-

mation nor quote any other analyses of the piece.

Four-Channel Tape and CD-ROM (4-ch)

Roberto Doati kindly provided me with a CD-ROM

with a recent digitization of the four-channel tape,

dated September 1988, that he used for the analysis of the piece (Doati 1988). Figure 3 shows the tape

box. Mr. Chowning personally sent him the analog tape (Doati 2005). The text on the box was written

by Mr. Chowning himself and testifies to the pres-

ence of four sounds (Test Tones) for testing right and

left, and front and back loudspeakers for performance

in concert (-6 dB, DBX type 1 noise reduction). The digitization itself was made in November 2002 by Alvise Vidolin from the analog tape. To do this, Mr. Vidolin " cooked" the four-channel Ampex Grand Master 456 tape (which had a dBx type 1 noise reduction) and then played it back on an OTARI four-channel recorder belonging to Centro

di Sonologia Computazionale (CSC) dell'Universita



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 3. Box of Mr. Doati's original four-channel tape.

Figure 3. Box of Mr. Doati's original four-channel tape. di Padova, connected to a Kyma system

di Padova, connected to a Kyma system (Capybara 320) equipped with an eight-channel, 24-bit audio


Wergo CD (WER)

The booklet of the Wergo CD indicates a 16 '5 7" du- ration. The track is stereophonic, which means that

a conversion from quadraphonic to stereophonic

format had been made. Nevertheless, the difference

of one minute between this and the four-channel

version is remarkable, indicating the presence of other important processes during CD mastering

that must be studied.

The liner notes by Johannes Goebel provide brief technical explications of the compositional tech-

nique: the " original quadraphonic version utilized

12 bits, two different sampling rates being used to

accommodate the enormous amount of data on the

magnetic disc-packs available at that time. The

original sound-data was processed by sampling-rate conversion and digital mixes to achieve the stereo version presented on this CD." The master tape of the CD "was made directly from the computer sys- tem at CCRMA which generated and stored the sound data to digital format. No analog recording was involved at any stage of the production and ed-

iting process" (Chowning 1988).

There follows an interview with John Chowning

where the composer explains the genesis of the piece, his work with Andy Moorer - who showed

him how to write a program in SAIL, and the fact

that all was absolutely deterministic. "There is not

a random number generator in that piece, which is

not true of Phone or Sabelithe or Turenas" (Chown-

ing 1988). Mr. Chowning concludes that the "idea of Stria was initially an aural phenomenon, but then it became an intellectual one which led my ear further than I possibly could have heard."

The IRCAM Tape (IR-ta)

This is a problematic source. Marc Battier kindly

gave to me this tape, the cover of which is shown in

Figure 4, which originally belonged to the IRCAM

archives. Unfortunately, I have not been able to dig- itize the tape yet. Because it is thirty years old and the magnetic surface is deteriorating, it requires a

professional procedure with several cookings before

the digitization. In any case, a short listening of the very initial seconds of the tape (necessary to under-

stand the physical condition of the tape) made it

possible to identify the beginning of some 440-Hz sinusoids used for testing the loudspeakers. This means that the tape could be quadraphonic, and

maybe it is a copy of the four-channel tape. This is

not immediately evident, as the box shows the in-

scription "quad?" Further research will be essential to complete the identification of this source. More-

over, it is not possible to deduce the duration of the

tape without listening to it. No dating is marked on the box, except for the date of realization of

the piece, which of course does not directly corre-

spond to the realization of the tape. Other technical characteristics of the tape, as indicated on the box, are the following: "green head, blue tail, 7.5 ips,

no-noise reduction."

IRCAM Digitization [IRdig)

This is a four-channel digital version on CD-ROM

provided by John Chowning. It probably is a copy of the digitization of the four-channel tape (4-ch).

However, the CD-ROM does not show any note on

the box, nor does the composer remember who

made the original digital version of the four-channel

tape of Stria. This must be verified. (According to Mr. Chowning, it was not he who made the digitiza- tion.) He is also "sure that it is not assembled from the digital files that Johannes [Goebel] used to make

46 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 4. The IRC AM four-

channel (I) tape.

Figure 4. The IRC AM four- channel (I) tape. the Wergo recording, because I would have

the Wergo recording, because I would have been in- volved." The sampling rate of the four tracks is listed at 44,100 Hz at 16-bit quantization, and the

overall duration is 17'26". As will be demonstrated

in the following section, the source had been digi- tized at a sampling rate of 48,000 Hz, causing a dif- ference in its duration easily adjustable by sampling

rate conversion.

John Cho wiring's Digital Data [JCdig)

These are sources scanned from paper printout of the

digital data and from various other documents, some

of them handwritten, used for the synthesis and the

calculation of the structure of the piece. They origi- nally belonged to Mr. Chowning's personal archive and are, since 2007, stored at Stanford University Archives in the John Chowning Papers. Among

them there are eight different files dating from 27

September 1977-3 October 1977. Each file shows

the date, the precise hour of the calculation job (e.g., 4:48, 14:33, 21:48, etc.) and the date of "printing," which is important to understand working hours of

the author. Each file shows the signature [PC, JC],

Figure 5. Header of JCdig

files (in this case, TO .MEMj.

which stands for "made by John Chowning and stocked in a personal area of digital memory at the

Artificial Intelligence Laboratory" (see Figure 5).

JCdig sources used for my research are of seven


1. The first, COMP . sai (one file) dated 3 Octo- ber 1977 and COMP . SAI (one file), dated 7

January 1978, consist of algorithms for the

generation of the piece (Chowning 1977d,

Chowning 1978a). 2. The second one (six modules named

TO .MEM, T286 .MEM, T466 .MEM, T610 .MEM,

T7 54G . MEM, end . mem) print a record of in-

put values requested by COMP . SAI and vari-

ous data that have been inserted, mem files

are records of data used for the generation of

each section of the piece (Chowning 1977a).

3. Next, one file (TO . SCR), dated 26 September 1977, is the "score" that the program

COMP . SAI generates. The file includes a se- ries of parameters for the synthesis of sound

(another type of file generated by the pro-

gram - with extension . rep - appears among

different data but was not used; this was

probably a report file for incidental errors, or

perhaps it is a lost file) (Chowning 1977c).

4. One file, dated 8 September 1977, is named

CPC4 . INS, with algorithms for the instru-

ment (Chowning 1977b).

5. Another file, dated 5 January 1978 and named CPC4 . INS, includes algorithms for reverberation (Chowning 1978b).



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

6. Another source includes a file with program

notes for the first performance, dated 5 Octo-

ber 1977 (Chowning 1977 1).

7. The next source includes two undated (most likely 1977) handwritten notes indicating various sampling rates for sections (Chown- ing, 1977(?)a) and the formal scheme of the

piece (Chowning 1977(?)b).

Other sources are included on the forthcoming Computer Music Journal Sound and Video Anthol-

ogy DVD that accompanies issue 31:4.

Comparative Evaluation and First Hypothesis

Philological investigation has passed three steps

thus far: analysis of the sources,- comparison among

them,- and comparison with external testimonies in-

cluding oral witnesses, electronic mail, and various

schemes and figures that John Chowning, the au-

thors of the other sources, and people involved in

the assembling process sent to me, in order to help my quest for the history and the circumstances in-

volving the realization of Stria. What follows are some reflections related particu- larly to the historical circumstances that bred the sources. A common point of reference is the overall duration of each source. I must emphasize the blurred digital-analog nature of the piece, which also characterizes each piece of computer music re- alized during this historical period: Namely, data

were digitally calculated but then fixed on an analog

recording. That justifies the congenital " impreci-

sion" within the digital-analog conversion (or copy from tape to tape) and, consequently, some inaccu- racies in analog audio sources (or their recent digiti- zation) and their comparison. For what concerns the TM source, Tod Machover

writes that the project of Stria was developed in 1976. This is the unique case where that date ap-

pears. John Chowning always declares, in his oral and written sources, he worked during the summer of 1977. But he also explains that the piece was on his mind for some years: in Berlin during 1974-1975 (according to Johannes Goebel's liner notes accom-

panying the Wergo recording, p. 12), "over a period

Figure 6. Stria program notes by John Chowning

(Chowning 1977 e).

of several years/' as he writes in the program notes

for the first performance in Figure 6 - and it may be

that IRCAM asked him to do the piece in 1975 or 1976 according to the composer. He also stresses

that the title came to his mind toward the end of its

realization (during Summer and Fall 1977). The TbMl source attempts to translate the piece onto traditional notation. The perceptual represen-

tation starts with long and high sounds. At 5 '20",

the first fall to low sounds occurs. From 8 '00" to

10' 10", there is a continuous thickening of spectra

with low sounds. This differs from Dodge and

Jerse's structure (DJ) which puts this climax at

lO'OO". This source does not reveal any low notes at 5'20". Probably, the DJ analysis was based on the digital data. So the difference in the spectral space ought to depend on the bandwidth of each modu- lated sound, which makes perceptive listening differ

from the pure computer data, but also from the dif-

ferent durations.

John Chowning explains that Mr. Mountain's

transcription (15'46" duration), is almost certainly

from copies of the Stria version that the composer

reassembled at IRCAM the day following the first performance, on 14 October 1977, and brought back

from Paris to CCRMA. This would relate the TbMl

analysis to the 4-ch analysis. Figure 7 shows the amplitude/time representa-

tion of the four-channel tape (4-ch) and the four-

loudspeaker arrangement (as deduced from RD). The amplitude-time representation of the piece lasts

48 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 7. Amplitude-time

representation of the four-

channel tape (4-chj. Chan-

nel 1 is front-right, 2 is

front-left, 3 is rear-left, and

4 is rear-right.

2 is front-left, 3 is rear-left, and 4 is rear-right. 16 '40". This happens because the

16 '40". This happens because the digitization of the

tape (and the tape itself) begins with 28 sec of four

groups of sinusoids at 440 Hz. The piece begins at 0'44" and ends at 16'32". (Here, the track contains a click followed by low hiss.) The piece is thus 15 '48" duration, which presents a quite negligible error of

2 sec compared to 15'46" cited in Doati (1988) and

probably caused by rounding the timing up.

Comparison between Mr. Mountain's graphical

score and the listening of Mr. Doati's four-channel tape suggests that this one and what Mr. Mountain analyzed are the same, as it appears from harmonic comparison through listening and from the dura- tions. Now, according to what John Chowning says about the Mountain analysis - namely, that that

was the version mixed the day following the first

performance - we can therefore establish that Mr. Doati's four-channel tape corresponds to the

"CCRMA version" (also confirmed by the com-

poser; Chowning 2005c).

Toby Mountain kindly provided me with extra notes he used for his analysis (TbM2; Mountain 1980). He says that all these notes were based on conversation with Mr. Chowning and his explana-

tion of the piece and that "almost none of those

notes represent original thinking on his own part"

(Mountain 2005). But they are very useful for my

analysis because they show some problems on num- ber, names, and durations of the sections that illu-

minate the compositional process.

IRCAM's digitization (IRdig) of the four-channel tape is problematic. The four audio files' properties show a sampling rate of 44,100 Hz with 16-bit quan- tization, and their duration is 17'26". Nevertheless, aural inspection reveals that the overall sound of the piece is lower (more or less a semitone) than the four-channel source. That means that the original

digitization had been made at a 48,000-Hz sampling

rate. Adjustment of the rate from 44,100 to 48,000

Hz and conversion again from 48,000 to 44,100 Hz lowers the duration to 16*01 ". Considering an easy

reject in the reading and differences in initial and fi-

nal silences of the tracks, this duration is compa- rable to 4-ch (15'46"). The waveforms are similar, as is demonstrated in Figure 8, even analyzed and

deeply zoomed more carefully with Cool Edit soft-

ware. IRdig does not start with four tests for the




This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 8. IRdig first channel

versus 4-ch first channel.

Figure 8. IRdig first channel versus 4-ch first channel. (To complete the historical scene, it is

(To complete the historical scene, it is worth not- ing that Pierre Boulez spoke about Stria for a Radio France broadcast lecture on 22-28 April 1980 at the

Theatre d'Orsay, Paris,- [Boulez 1980]. The excerpts

from Stria are played at the proper pitch. This

demonstrates that the IRdig source, the digitization of the analog tape, was made after 1980 when this cassette was made. The fact that he played only a few excerpts unfortunately does not permit us to deduce if this was the first long IRCAM version.) I can state that the IRCAM digitization (IRdig) does correspond to the four-channel source, or that concerts at IRCAM normally played the " false" ver- sion a half-tone lower in pitch. DJ (Dodge and Jerse's famous scheme) registers the longest time duration. Why does this scheme at- tach an 18-min duration to the piece? Mr. Chown- ing notes that this scheme is conceptual rather than

actual (Chowning 2004). He remembers having told

Charles Dodge that the duration was 18 '00", forget-

ting that he had shortened Stria in the process of re-

assembling. And he also remembers that the first performance was even more than 18'00" in duration,

which gives us again a sixth time duration. (This

would mean that the IRCAM version was closer to

18'00", rather than 15'00", as their Web site indi- cates.) Mr. Chowning says that Mr. Dodge asked him

to join him for a short time in a cafe after some

event somewhere on the East coast. I explained the basic scheme of Stria, the relationship of

the golden section to the spectra and pitch space, the formal division 610/987 (16'27" = 987) being the Golden Section, and how I com- posed it by developing a recursive procedure. He did not ask for any material. Their book was published in 1985. So they did not have the Wergo CD either, and I have no knowledge of

their having a copy of the tape, although they

may have. Perhaps they heard it at the ICMC at

UCSD (Chowning 2005c).

At IRCAM, it seems impossible to find any docu-

mentation, written or audio, of the first version or

any digital data left (as TM-LK informs). My re- search on the very first version (IRCAM's first per-

formance) stops here concerning audio sources.

50 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 9. Nominal version large-time divisions versus the actual version (Chown-


divisions versus the actual version (Chown- ing2004). Assembling Process My investigation on the assembling

Assembling Process

My investigation on the assembling focuses there- fore on 4-ch and WER audio sources. I have already pointed out that Mr. Doati's four-channel tape and

the Wergo CD differ by about 1 min; this is a very

important divergence, and their waveforms are in

some points different. It is important to note the

electroacoustic community has never been both-

ered by this difference in timing.

At the beginning of my research, Mr. Chowning

himself explained that such differences were caused by the new mixing realized for the CD, but he did not recall the precise circumstances that caused this important change. Nevertheless, he initially pro- vided me with some Excel files that tried to explain all the passages. Mr. Chowning also declared that

some differences from his ideal sketches (perhaps the

IRC AM version or some previous sketches?) and the

actual piece of music exist. The nominal shape is

what was in my head (classical geometric divi- sion), but music, of course, is not sensed only in metered time (and certainly not Stria). There are so many factors such as density, loudness, pitch height, duration, etc., of sound, so what was composed is what was in my ear relative to

the abstract geometric model. Where exactly

the climax occurs is still a mystery to me as it is a region. The nature of the piece does not re- ally suggest points. I only know that it is before

8'25", and at 8'25" there is a kind of sectional

closure (Chowning 2004).

Figure 9 shows what Mr. Chowning intended, in

2004, as a nominal version large time divisions of the

piece and an actual one (Chowning 2004). The nom-

inal version is the composer's first intention, that is

the classical Golden Section division. (Note that

1.6180 is the actual ratio of the values 987:610:377.) Nevertheless, no duration of my documents matches the overall duration of this nominal version (16'27"). Mr. Chowning then indicates what he intended to be the "actual " version. This is a bit longer (16'54"). The proportions are modified but closer to the Wergo CD duration, which is 16'56". A coda is added from 8 '30", and the sections in some cases

follow the Golden Mean (830:505 = 1.643). It is also

clear, at this point, that Dodge's and Jerse's diagram (DJ) has totally different proportions (18'00") rela-

tive to all other versions. If one takes this schema

for granted, WER could be the version that better

satisfies the actual version, but this invalidated by

the next considerations.



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 10. Assembling of

Stria as recorded on the

Wergo CD versus nominal

and actual versions

(Chowning 2004).

This document differs from the one published by

Bruno Bossis in 2005, which shows an original (nominal) version with two blocks (630" - 389") and

an actual version made by two sections (506" -

324") and a coda (Bossis 2005, p. 101). Trying to ex-

plain these discrepancies, in another later Excel file, Mr. Chowning compares the Wergo audio version with the four-channel tape and the actual version

(see Figure 10), trying to rebuild his own composi-

tional process. "I have made myself go back and try

and remember what I did in 1977 and 1988 when the

CD was produced. This is what I have reconstructed

from the * . mem files and the CD recording. I re-

member now that I truncated the second section, but never modified the data" (Chowning 2004). The

nominal and actual versions have here the same du-

ration (16'59" = 1,019 sec): the Wergo version lasts 17'00" (1,020 seconds), quite close to the audio Wergo version (WER), which lasts 16'56". This also

slightly differs from the actual version in Figure 10

(1,014 sec, 16'54"). Figure 10 is also presented in Bossis's article (Bossis 2005, p. 102). These were Chowning's first explanations and memories, but real durations and assembling pro-

cesses need to be verified. It is useful therefore to

compare ( 1 ) the computer algorithms, input data,

and digital scores (JCdig) with (2) the Excel files

with schemes and numerical data related to the

timing and (3) the audio files. According to Figure 10, Stria, in the Wergo version, is shaped in six sections, some of which

are slightly overlapping. These are TO . mem,

T286.MEM, T466.MEM, T610.MEM, T754G.MEM,

and end . mem. The number after T indicates the

time of initialization.

Each section corresponds effectively to the data

listings JCdig, with the . MEM extension represent-

ing the input to the COMP . SAI procedures. TO . MEM

was sampled at 25.6 kHz, T286 .MEM at 12.8 kHz (an important note in Figure 10 says " truncated"), T466 .MEM at 25.6 kHz, T610 .MEM at 12.8 kHz,

T7 54G . MEM at 25.6 kHz, and END . MEM at 25.6 kHz.

(The same sampling rates are shown in Figure 18, handwritten by the composer.)

Table 2 shows different initialization time of the

six sections (Tx . mem), as indicated in the JCdig

computer data. The end of the listing of each . MEM

file shows the complete value (in sec) of each sec- tion ("Duration"). Each file also indicates the tim-

ing of the printed output; with the exception of

52 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Table 2. Mr. Chowning's . mem Files

Computer Data

Section Duration File last written

TO .MEM 166 (2'46") 4:48, 2 October 1977

T286 .MEM 325 (5'25") 2:37, 3 October 1977

T466 .MEM 144 (2'24M) 7:21, 27 September 1977 T610 .MEM 144 (2'24") 17:53, 29 September 1977 T754G.MEM 234 (3'54") 0:41, 3 October 1977 END. MEM 145 (2'25") 4:45, 3 October 1977

T610 calculated at 17:53 (i.e., 5:53 pm), the other sections were realized at night. Mr. Chowning

clearly demonstrated the late-night work ethic at-

tributed to many computer music pioneers with

"heavy eyelids/7 who were forced to work at night

owing to policies limiting the use of the PDP-10 for music and the very long calculation time owing to

time-sharing. The synthesis instruments were in the form that

was established by Max Mathews (1969), which

consisted of unit generators that synthesised and controlled the acoustic signals. COMP . SAI included

the algorithms and generates three different files:

x . SCR (the score), x . REP (never used), and x . MEM.

The file TO . SCR is the only complete set of instru-

ment and reverberator data. The . SCR file consisted

of the time-ordered parameters that were passed to

the synthesis algorithms. The parameters of the

. SCR file were the start times and durations of the

instruments, the frequencies of the FM oscillators,

the name of the functions used for amplitude en-

velopes and frequency skew, the angular and radial position of the instrument output, etc. (Chowning


It is clear from Table 2 that initialization times

compared to durations do not always coincide (e.g.,

TO + 166 sec * 286). To try to explain that, it is help-

ful to compare all the data collected. Table 3 col- lates all durations marked in Figure 10 and Table 2.

On the left are the data of the Excel file represented

in Table 2 (init time values are recollected by John

Chowning). On the right are the timing listings as set up in the computer data JCdig.

end . mem was the last file to be calculated, i.e.,

digitally generated but not evidently conceived; it

was 4:45 am, ten days before the first performance. Mr. Chowning adds: "In fact it was one of the first

sections that I generated; as I remember it was origi- nally TO. After much consideration, I decided that it was the perfect ending according to my evolving

conception of the piece and I renamed it end . mem

( . SCR, . rep)." This is also confirmed by the rela-

tively few input parameters for an earlier form of

COMP. SAI (Chowning 1977d, 2006, 2007).

The x in the T . mem files (0, 286, 466, etc.) stands

for the time initialization (in seconds) and uses Fi- bonacci numbers. Fibonacci numbers (beyond the synthesis of the frequency space) are used in Stria to determine start times, attack durations (the time within which elements can begin within an event), and overall durations. However, Mr. Chowning ex- plains that the values are not always simple Fi- bonacci numbers, but additions of the numbers

taken from the series, for example 466 = 89 + 144 +

233 (Chowning 2005c).

Table 3 shows that the x timings of the sections assembled in the Wergo version ( WER) and those synthesised by the computer data correspond. T745g WER (left) corresponds to T754G; an error made by Mr. Chowning in the compilation of the Excel file reversed the numbers, but the sections are

the same (Chowning 2005c).

Computer data listings and x values of the . MEM files bear traces of the removal or change of some

sections: This happens for the gap between the end of TO (2'46") and the beginning of T286 (3 '26"). This

could mean two things: ( 1 ) that the x . MEM numbers

cannot always stand for the time initialization of

each file - in fact, the time initialization of the pure

x value would be O'OO", 4'46" (T286), 7'46" (T466), lO'lO" (T610), 12'34" (T754G), ? (end); or (2) that the composer originally created some other sections

and later deleted them, which is more probable and

will be demonstrated.

All this reveals that Mr. Chowning's original project was different from its musical realization. Above all, he conceived some other sections that must stand between the ones we know, for example

between TO and T2 8 6, to cover the gap I have

pointed out. The score bears out this theory. The procedure inharm - shown in Figure 1 1 and which

is part of the COMP . SAI file - refers to some sec-



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Table 3. Comparison Between Wergo Assembling (as Hypothesized by John Chowning) and JCdig

WERGO assembling Computer data

Section Init time End time Duration Init time Duration File last written

TO. MEM 0 166(2'46") 166(2'46") 0 166 (2'46") 4:48, 2 Oct. 77 T286.MEM* 163 (2 '43") 369(6'9M) 206(3'26") 206(3'26") 325 (5'25") 2:37, 3 Oct. 77 T466.MEM 365 (6I5M) 509 (8'29") 144(2'2411) 466 (7'4611) 144(2I24") 7:21, 27 Sept. 77 T610.MEM 506(8'26") 650(10'50") 144 (2'24M) 650(10'50") 144(2'24") 17:53, 29 Sept. 77 T745g.MEM 648 (10'48M) 882 (14'42") 234(3'54M) 754(12'34") 234(3'54") 0:41, 3 Oct. 77

END. MEM 875 (14'35M) 1020(17'00M) 145 (2'25") 145 (2'25"j 4:45, 3 Oct. 77

*The original duration (325") was truncated for the actual CD (206")

tions named T4 6 6 A and T3 7 7 . The A after T4 6 6 was

just to identify it as one of some number of ver-

sions, e.g. T466, T466A, T466B, that he tried for

that section,- this was similar for T7 54G . MEM

(Chowning 2005c).

In May 2005, Mr. Chowning found another mean-

ingful handwritten document (undated) showing that the original plan (see Figure 12) included an- other section named T163, which would have filled

the gap between TO and T286. In fact, the last start time of TO .mem is 159.682", and its latest end time

is 166" (data taken from the Excel file and TO . mem).

T163 would have slightly overlapped, beginning at 163". This Stria version is shaped in seven sections,

which include TO, T163, T286, T466, T610, T754F,

and end. Even the F of T754 is the only occurrence,-

the other important detail shown in Figure 12 is the overall duration: 987", that is, 16'27". Mr. Chowning finally decided not to include

T163 or T377 (Chowning 2005c). Toby Mountain's

sketches [TbM2; Mountain 1980) are also revealing

of this process. Mr. Mountain explains: "I think

John gave me a copy of the computer printout, but I don't think I used it. My notes with the different

times for the sections were based on what he told

me" (Mountain 2005). One of Mr. Mountain's notes

(see Figure 13) refers to six sections in the following

order: TO, T166, T466, T610, T754, and END. We can see that there certainly was a section named T166 (I have decided to accept this discrepancy - T166 in- stead of Tl 63 of Figure 12, which could be simply a later section with more refined instructions), but in this case, T286 is lacking, as it was an alternate sec- tion. The note is useful though, because it indicates

Figure 11. Inharm procedure

(JCdig; Chowning 1977 d).

the duration of T166 (268 sec) and the duration of the single events within each section. Stimulated by these observations, the composer

also remembers that the first performance still in-

cluded the T166 section, which was a very static

54 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 12. Handwritten document by John Chown- ing (Chowning 1911 (l)b).

document by John Chown- ing (Chowning 1911 (l)b). section. T754 was the original end, but before

section. T754 was the original end, but before leav-

ing for the first performance, he added the END sec-

tion, both of which end on C-sharp7. Actually, he

had realized end early in composing the piece (JCdig indicates it was the last to be calculated for the definitive synthesis of the sections), which is

why the . MEM data is shorter than the others. The

program grew as Mr. Chowning developed the piece

(Chowning 2005c).

Following the first performance, the composer truncated T166 and maintained six sections: TO,

T286, T466, T610, T754G, and END. The truncation

is quoted also by another note by Toby Mountain (in a file called note . 03 ) which reads: "Total time

15:46, 946 sec (original 987"), Golden section

584 - 9:44 climax, eliminated one section"

(Mountain 1980). Mr. Mountain's quotation - 987

sec, or 16'27" - correspond to the version with

seven sections.

At this point, I help my investigation by the lis-

tening analysis and corresponding amplitude-time representation. The comparison between 4-ch and WER audio versions was made by listening carefully to the sound files. In this way it has been possible to

locate where each section begins, how it sounds, and how it is mixed with the following section. The most evident element (in the time-amplitude repre- sentation) is that - as I have already remarked in



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 13. Toby Mountain's note . 07 sketch (Moun- tain 1980).

13. Toby Mountain's note . 07 sketch (Moun- tain 1980). 56 Computer Music Journal This content

56 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 14. Four-channel

tape versus Wergo version.

Figure 14. Four-channel tape versus Wergo version. the different time durations - the four-channel tape is

the different time durations - the four-channel tape is shorter than the Wergo version by one minute

(see Figure 14). The assembling for the Wergo CD was made by Johannes Goebel, a composer/producer/technician,

on behalf of the Schott firm, of which Wergo is a

part. According to his liner notes, Mr. Goebel made the CD in 1988 from the original sound files that

had been transferred from the mass storage disk

(perhaps Librascope, as Mr. Chowning remembers) and kept on digital tapes (Chowning 2005c).

The reason for the longer Wergo version's dura-

tion dates back to the collaboration between Mr.

Goebel and Mr. Chowning. At the beginning of my research, Mr. Chowning recalled that Mr. Goebel took with him two alternative sections to the origi- nal sections (four-channel version): (1) one following

2'43", "that I had originally considered" (Chowning

2004); (2) an extended section before the end that Chowning had originally truncated when he recorded it onto tape. (At CCRMA, composers and staff never discarded sampled data, as it took hours

to compute; Chowning 2005c.) When Mr. Goebel

was doing this, it was a very busy time at CCRMA, so Mr. Chowning just controlled and listened to the

beginning section and loud section before 8 '25" to

confirm levels. He did not realize that it was not the

same version as the original before Johannes Goebel returned to Germany. The Wergo CD was produced with that version (Chowning 2005c). Comparing a listening to the amplitude-time representation, it is evident that (1) T286 lasts 252 sec in 4-ch; (2) T286

lasts 204 sec in WER.

If we consider normal errors owing to imprecise selection of the audio files and the overlapping of

the sections and we take into account data marked

in Table 3, we can assume that ( 1 ) T2 8 6 in WER

lasts 206 sec (204 sec in my listening is an accept- able error); (2) in 4-ch, T286 lasts 252 sec, but fCdig

T2 8 6 . MEM indicates 325 sec. It is a remarkable dis-

crepancy that apparently hides three steps in the history of T2 86: synthesis (325" duration), first

probable truncation (252"), and third truncation (WER). The composer later has explained that T2 86

was truncated the first time because T286 was simi-

lar to the following T466 (Chowning 2005c). The

last four events of both T2 8 6 and T4 6 6 have the

same durations and attack durations, with only a different base frequency (one pseudo-octave higher in T466). This had led him to replace all of the last



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 15. Discontinuity at

6'29" (389").

Figure 15. Discontinuity at 6'29" (389"). part of T2 86 with T466 (Chowning 2006). Concern- ing

part of T2 86 with T466 (Chowning 2006). Concern- ing T754G, my listening reveals that (1) T7 54G in 4-ch lasts 1 1 1 sec; (2) T754G in WER lasts 230 sec.

Because sections that Mr. Goebel did use were the

original sections converted to different tapes from

digital data, it is clear that: (1) T754G lasts 234 sec,

as marked in Table 3, in both WER and fCdig; (2) T754G, with a duration of 234 sec, is the one origi-

nally calculated since the first writing of T7 54 . MEM

section,- (3) the short T754G used in 4-ch is a version

cut down at a certain point of the realization (before

or after the first performance at IRCAM).

The comparison pointed out in Figure 14 has helped Mr. Chowning remember that some of the events happened in 1977 and suggested to him why the Wergo version was cut by Johannes Goebel. (The truncated part of T286 is marked with a curve in the graph.) Mr. Chowning remembers that in the

CCRMA version (4-ch), there was a "discontinuity"

in the D/A conversion of the data at 6'29" (389") from the original computation that was unintended. This caused a sudden change of timbre and a conse-

quent click: "[T]he PDP-10 burped!" (Chowning

2005c). This imperfection in the computation

emerges slightly, but very clearly indeed, in the au-

dio source and with a zoom of the sonogram. The

discontinuity is marked by a spot in the Figure 14,

and it is zoomed in Figure 15. We can see that the

constant flowing of the FM spectra is broken by the


My listening confirms that T286 was already trun-

cated (252 sec instead of 325 sec); in WER, the audio

is truncated exactly at that point (6'29") with a fade out to the following section. Owing to the large

amount of the computational effort involved at that

time, Mr. Chowning did not re-compute the section to eliminate this problem. He rather learned to ac-

cept it "as one does a birth mark or beauty mark on

one's skin

noticeable but of no substantive con-

sequence" (Chowning 2005c). This comment also

makes clear that both IRC AM and CCRMA [4-ch) versions contained the discontinuity.

What Mr. Chowning sa^s about the truncation is that he cannot "remember any conversation with

Johannes about that, but that the truncation hap-

pens just before the 'click7 makes me suspect that he/we decided to truncate before" (Chowning 2005c). Nevertheless, the truncated part lasts from 6' 15" to 6'57"- that is, 43 sec. This is not actually

the same durations Chowning declares in the Excel

file (Figure 10 and Table 3), which lists a truncation

of about 2 min (325 - 206 = 1 19 sec). The Excel file

could be considered a hypothetical reconstruction

by Mr. Chowning rather than a philological recon-

struction of the process. (This is valid also for the

image used in Mr. Bossis' article.)

The "faulty," imperfect, and therefore fascinating

four-channel version is the version Mr. Chowning now uses to play during the concerts. (I have person-

ally verified this fact during a concert given at the

Bourges Festival of Electroacoustic Music during June 2005.) Thus, we should emphasize that the

Stria version that audiences know from live perfor-

mance with spatialization has a discontinuity in the D/A conversion at 6 '29".

WER also contains an extended fifth T754G sec-

tion that is completely truncated in 4-ch. The fact that the Wergo CD was made from the original

sound files that had been transferred from the mass

storage disk make me suspect that the original T754G had probably been considered too long by the composer for the CCRMA version and, maybe, for the IRCAM version, but that the truncation the

CD was neglected, for reasons that I leave to further


58 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Another discrepancy emerges in the overlapping of some sections. What Johannes Goebel says is that

after listening to Stria many times before putting

the CD together, he saw that there was never a pause in it anywhere (Goebel 2005). WER does not contain silence, but the 4-ch version almost dimin-

ishes to silence between the first section TO and

the second, T286. In the Wergo version, the two

sections overlap for 3 sec. (Table 3, left, shows the

time overlapping.) The CCRMA version (4-ch)

fades to a very feeble sound; during the first listen-

ing, I thought there was a silence of 2 sec. As Mr.

Chowning says, "This difference was not preserved

in the Wergo version. This was an important break or caesura in my conception of the piece" (Chown- ing 2005c).

Another important - this time functional - differ-

ence between the CCRMA version (4-ch) and the

Wergo ( WER) version relates to the dynamic range. John Chowning informed me:

When I saw your two-channel/four-channel sig- nal comparison [Figure 14], I remembered an important discussion that Johannes and I had having to do with dynamic range. The loudest

parts of the two versions are at about the same level, but the soft parts of the Wergo CD are much louder. Johannes felt that if the begin-

ning was as soft as I had composed it, the lis- tener would turn up the volume, thinking it

was under-recorded and reduce it later at the

loud part. He felt that a compromise was better than distorted or interrupted listening (Chown- ing 2005).

That was a practical adjustment necessary for do- mestic listening. After this discussion, Mr. Chown-

ing remembers he only had time

to listen to and agree on the beginning and the loud section. Johannes, also extremely busy, re- turned to Germany and I did not realize that the Wergo version and the tape that I was playing at

concerts are not the same. The Wergo CD of

Stria is just under 17', while the 4-ch tape that

I assembled and mixed with the sectional over-

laps at IRCAM before and again right after the

first performance, is 15 '46" (Chowning 2005c).

Ultimate Description of the Assembling Process

It is now possible to trace a first attempt to recon-

struct the process of Stria's assembling. The analysis of different sources would, in many cases, have been

insufficient for my purposes. It has been helped by

oral communications, which in some cases, still

need further research. The actual investigation

leads me to outline the stages in the history of the

mixing, listed in Table 4. 1 have submitted my con- clusions to John Chowning, who has personally added some other steps he remembered. These new stages, which also need to be further validated, are

within brackets. (Step 3 is particularly interesting, because it cannot be verified by any of the sources.)

These results show the history of the synthesis of

sections and the various stages of assembling. Four

stages emerge: ( 1 ) a complex compositional process of sections 1-5; (2) the IRCAM conjectural version

before the premiere (future research hopes to find the tape of this very first IRCAM version); (3) the

CCRMA version made the day following the pre-

miere,- and (4) the Wergo recording.

Further investigation could bring to light other

sources or witnesses that could add other details

and confirm or contradict my conclusions. I wish to underline that the philological investigation has

stressed the important role of the collaboration be- tween the composer/producer/technician Johannes Goebel and John Chowning. A collaboration caus-

ing an interesting double identity of Stria that has never perturbed the fortune of the piece, its beauty

and the listeners' appreciation. Criticism based on Table 1 and what emerges from Table 4 emphasizes that the only real auto-

graph among these sources is what I have called

fCdig. An autograph is a "manuscript" by the au-

thor. These are printed sources from the original digital data used for the synthesis or other hand-

written documents. Deleted sections (Table 4, step

1, T166, longer and shorter sections) are the so- called sketches. Some of them unfortunately have been lost but their traces remain in the fCdig source. The Wergo version appears to be an author- ized edition of the piece, realized under the control of Mr. Chowning. But the fact that the penultimate section shows a difference in timing could make



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Table 4. History of the Various Stages of Assembling

Phase Date Number of sections Comments

1 1977 ? First experiments included TO, a shorter section END, T163, T377, T466A,

inharm (Figure 11), and a shorter section than the others, with the proce-

dure event2 in a less-developed state (the composer later renamed it

END. MEM; JCdig)

2 1977 4 Sections T166-T286 (325")-T610- T7 54G (long version, 234")

3 1977 5 The composer creates T466 (144") and decides to truncate T286 (the first trun-

cation is faded out, but Mr. Chowning keeps the data: 252" instead of 325")

4 1977 6 sections Alternative sections, according to Mr. Mountain's sketches (Figure 13): TO

(166")-T166 (268")-T466 (144")-T610 (144")-T754 (version 234" )-END (145",

without T286) (Mountain 1980, note. 07)

Duration: absolute duration (without overlapping) 1101", i.e., 18 "21" (per-

haps this corresponds to Dodge and Jerse's scheme); Mountain 1980 (note . 03) indicates 16'27" (987"); future research is needed

5 1977 6 sections Nominal version: hypothetical reconstruction by John Chowning adapted to

the WER version, before the philological investigation (Figure 10, June 2004)

TO (166")-T286 (252" version, with discontinuity)-T466 (144")-T610 (144")-T754G (234")-END (145")

Duration (with overlapping): 1019", i.e., 16'59"

6 1977 7 sections IRCAM version, first performance (Figure 12)

IRCAM version Duration in Figure 12: 987", i.e., 16'27"

TO (166")-T163/6 (268")-T286 (252" version with discontinuity )-T4 6 6


Absolute duration without overlapping: 1230" (20'30" if T7 54G is 111") or

1353" (22'33" if T7 54G is 234")

7 1977-1978 6 sections Following the first performance, Mr. Chowning manually cut the whole

CCRMA version section tape of Tl 6 6 (static section).

Sections TO (166")-T286 (252" version with discontinuity)-T466 (144")-T610 (144")-T7 54G (short version, 111")-END (145")

Duration: 15'46"

8 1988 6 sections T286 third truncation (206"): Mr. Chowning truncated the discontinuity

Wergo version T754G (long version, 234")

Compromise in dynamic range

Duration: 16'57"

one deduce that it is a version with the so-called ed- itorial integrations, which has become authorized

in the course of the years. Variant criticism has

identified two alternative variants. Variants exist

when two or more versions co-exist, and their au- thor does not choose between them.

It is also meaningful to show the last Excel file

(see Figure 16) John Chowning sent to me, present-

ing what he recollects after my and his research work.

The version for the first performance still remains

the most problematic version, because its timing conflict with my reconstructions and remain ob- scure. Further investigation could adjoin new infor- mation concerning the various stages of assembling and an additional aspect not considered in my re- search, which is the control and the accuracy of

60 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Figure 16. A comparison of

the IRC AM, Wergo, and

CCRMA versions made by

John Chowning (2006).

Figure 1 7. Excel data with

durations and overlapping

of the sections, as recon- structed by John Chown-

ing (2007). Note: In this

figure, John Chowning re-

duced the durations of sec- tions TO, T286, T754, and

END, according to their be- gin time of 1 sec rather

than 0 sec. (The offset was

to allow editing out a click

produced on the tape by

the DACs [Chowning

2007].) For the purpose of

my research, I decided to consider only the pure data

shown in JCdig.

I decided to consider only the pure data shown in JCdig. Figure 16 Figure 1 7

Figure 16

to consider only the pure data shown in JCdig. Figure 16 Figure 1 7 the Fibonacci

Figure 1 7

the Fibonacci series applied to the various stages of


The analysis by synthesis and re-synthesis of the piece is a fundamental counterpart to my conclu-

sions. Olivier Baudouin and Kevin Dahan have

faced, among other aspects, the problem of dura-

tions and exact timings of the overlapping sections

(see Baudouin 2007 and Dahan 2007 in this issue),

as shown in Figure 17. It must be emphasized that their works, super-



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

vised by the composer, have led to a reconstruction of the piece that matches the CCRMA four-channel

version (Chowning 2007). That allows me to state

that the CCRMA version corresponds to the best Stria version. The re-synthesis corresponds partly to the authorial level (Caraci Vela 2005, p. 221), namely, the moment in which authors establish their own true versions. This is the revision by the author himself, where he fixes a new and definitive

version of his work. On the other hand, it is not en-

tirely a revision of his own, since the synthesis is made by Mr. Baudouin and Mr. Dahan. This fact

could involve other consequences, musicologically

speaking, related to the interpretation, the identity

of music and, maybe, possible "new" versions.

Another important and problematic aspect Mr.

Baudouin and Mr. Dahan have encountered is the

reverberation. One last source Mr. Chowning pro-

vided was Figure 18 (Chowning 2007). To complete

the list of sources concerning the piece, I think it is

worth presenting it, together with Mr. Chowning's


Trying to remember how I modulated using

zdelay, I realized something important. In the

collection of notes that I have, there is this one [Figure 18] where I recorded the sampling rates that I used for each of the sections. The expense (in time) of these computations was so great that I economized. However, I am sure that I did not change the delay lengths (in samples)

accordingly. So, it is an artifact; the reverbera-

tion time in T2 8 6 and T6 1 0 is twice as long as

in the other sections (Chowning 2007)!

Following Mr. Baudouin's and Mr. Dahan's pro-

cess of re-synthesis, it has been even clearer that the

process of composition, even for one computer- music piece that is an intermediate point between parameter by parameter composition and automatic

composition, pursues complex mechanisms where

the machine control and determinism charmingly meets the human imprecision in the process of artistic creation. That is why I would like to con- clude quoting John Chowning himself: "[T]here are times that the composer, even in his most rigorous mode (especially in those days when forced late

nights, long computations, fatigue, and perfor-

Figure 18. Sampling rates of the sections (Chowning 1977 (l)a).

18. Sampling rates of the sections (Chowning 1977 (l)a). mance deadlines were not very compatible), is

mance deadlines were not very compatible), is led by his ear to accept what has been done, as an objet

trouve" (Chowning 2007).


I wish to thank all the people who have made this paper possible: Marc Battier, Massimo Borghesan,

Sergio Canazza, Giovanni De Poli, Roberto Doati,

Sergio Durante, Johannes Goebel, Lev Koblyakov, Toby Mountain, Jean-Claude Risset, Alvise Vidolin, and David Wessel. A special thanks to John Chown-

ing for his interest, kindness, and tolerance in an-

swering the long series of electronic mails, international phone calls, and meetings we have ex- changed in these years and for "cleaning up" my En- glish. It has been an honor to share with him the deep knowledge of his musical world.

62 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to


Barriere, J.-B. 1990. Pour une esthetique de la musique

avec ordinateur. These de doctorat. Univ. de Paris I -


Baudouin, O. 2006. Probleme de r analyse des musiques concues pour supports enregistres. Comparaison de deux oeuvres : Artikulation (1958) de Gyorgy Ligeti et Stria (1977) de John Chowning. Universite de Rouen, Faculte des Lettres et Sciences Humaines, Departe- ment de musicologie. Memoire pour l'obtention du

Master II.

Baudouin, O. 2007. "A Reconstruction of Stria." Com-

puter Music Journal 3 1(3): 75-81.

Borghesan, M. 2005. Porting di Stria di John Chowning in

ambient e PC. Unpublished thesis, Universita di Padova, Facolta di ingegneria.

Bossis, B. 2005. "Stria de John Chowning ou l'oxymoron musical: du nombre d'or comme poetique." In P. A. Cas-

tanet et al., eds. John Chowning: Portraits polychromes.

Paris: INA/GRM and Michel de Maule, pp. 87-115. Boulez, P. 1980. Pierre Boulez presente: Stria (1976); ex- traits I John Chowning. Kontakte (I960); extraits I Karlheinz Stockhausen. Arcus (1978); extraits I York

Holler. Audio cassette, IRCAM-Radio France Serie II /

Cassette 3 (Materiau et invention musicale, Vol. 3.). (Audio recording of concerts presented at IRCAM and the Ensemble InterContemporain, 22-28 April 1980,

recorded at the Theatre d'Orsay, Paris).

Caraci Vela, M. 2005. La filologia musicale. Istituzioni, storia, strumenti critici (I - Fondamenti storici e

metodologici della filologia musicale). Lucca: Libreria

Musicale Italiana.

Castanet, P. A., et al., eds. 2005. John Chowning: Portraits

polychromes. Paris: INA/GRM and Michel de Maule. Chowning, J. 1973. "The Synthesis of Complex Audio Spectra by Means of Frequency Modulation/' Journal of the Audio Engineering Society 21:526-534.

Chowning, J. 1977a. TO .MEM, T286 .MEM, T466 .MEM,

T6 1 0 . MEM, T7 5 4G . MEM, and END . MEM. Computer file

printouts. John Chowning Papers, Stanford University


Chowning, J. 1977b. CPC4 . INS. Computer file printout dated 8 September. John Chowning Papers, Stanford

University Archives.

Chowning, J. 1977c. TO . SCR. Computer file printout

dated 26 September. John Chowning Papers, Stanford

University Archives.

Chowning, J. 1977d. COMP . SAI. Computer file printout dated 3 October. John Chowning Papers, Stanford Uni- versity Archives.

Chowning, J. 1977e. Notesl977. Computer file printout

dated 5 October. John Chowning Papers, Stanford Uni- versity Archives.

Chowning, J. 1977(?)a. StriaSecsSR.pdf. Computer

scan of undated (most likely 1977) handwritten note in- dicating various sampling rates for sections. John

Chowning Papers, Stanford University Archives. Chowning, J. 1977(?)b. PlanLarge (version with 7 sec-

tions). Undated (most likely 1977) handwritten note in-

dicating various sampling rates for sections. John

Chowning Papers, Stanford University Archives.

Chowning, J. 1978a. COMP . SAI. Computer file printout dated 7 January. John Chowning Papers, Stanford Uni- versity Archives. Chowning, J. 1978b. CPC4 . INS. Computer file printout dated 5 January. John Chowning Papers, Stanford Uni- versity Archives. Chowning, J. 1988. John Chowning: Turenas, Stria,

Phone, Sabelithe. Audio compact disc, with booklet

containing an interview recorded on 8 May 1987, tran- scribed and edited by J. Goebel. Mainz: Schott/Wergo

Music Media, WER 2012-50.

Chowning, J. 2004. Personal communication, May, June, and August. Chowning, J. 2005a. "Composer le son lui-meme." In

P. A. Castanet, et al., eds. John Chowning: Portraits

polychromes. Paris: INA/GRM and Michel de Maule,

pp. 25-30.

Chowning, J. 2005b. Liner notes accompanying audio recording entitled OHM+: the early gurus of Electronic

Music 1948-1980. Audio compact discs and DVD. Ellip-

sis Arts CD 3693/DVD 3694.

Chowning, J. 2005c. Personal communication, March,

April, June, and September.

Chowning, J. 2006. Personal communication, June, July,

and September. Chowning, J. 2007. Personal communication, February-


Chowning, J., and D. Keislar. 2007. Personal com-


Dahan, K. 2001. Systemes dynamiques dans la creation

musicale assistee par ordinateur. MPhil thesis, Univer- site de Paris VIII, Departement de Musicologie. Dahan, K. 2007. "Surface Tensions: Dynamics of Stria."

Computer Music Journal 31(3 ):65-74. De Benedictis, A. I. 2004. "Scrittura e supporti nel Nove- cento: alcune riflessioni e un esempio (Ausstrahlung di

Bruno Maderna)." La scrittura come rappresentazione del

pensiero musicale (G. Borio ed.). Pisa: ETS, pp. 237-291. Doati, R. 1988. "Analysis of Stria by John Chowning." Unpublished article.



This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to

Doati, R. 2005. Personal communication, April. Dodge, C, and T. A. Jerse. 1985. Computer Music: Syn- thesis, Composition, and Performance. New York:


Gayou, E. 2005. "Entretien entre John Chowning et Eve-

lyne Gayou/7 In P. A. Castanet et al., eds. John Chown-

ing: Portraits polychromes. Paris: INA/GRM and

Michel de Maule, pp. 7-24.

Goebel, J. 2005. Personal communication, April.

IRCAM. 2005. "John Chowning: Stria/7

textes/cOOOOOO 1 8/n00000402/index.html (accessed

28 February 2007).

Koblyakov, L. 2005. Personal communication, September.

Lorrain, D. 1980. "Analyse de la bande magnetique de

l7oeuvre de Jean-Claude Risset77 Inharmonique. Rap-

ports IRCAM 26/80. Machover, T. 1984. "The Realization of Stria: The Devel-

opment of Human Singing Voice Synthesis Using Fre- quency Modulation.77 IRCAM Archives, A collection of Musical documentation (Internal documentation of re- search), pp. 17-19. Unpublished document. Mathews, M. 1969. The Technology of Computer Music.

Cambridge, Massachusetts: MIT Press.

Means, L. 2005. "Interview with John Chowning by Loren Means.77 YLEM Journal: Artists Using Science and Technology 25(6-8). Available on-line at http://www (accessed

28 February 2007).

Meneghini, M. 2003. "Stria by John Chowning: Analysis of the Compositional Process.77 Proceedings of the XIV

Colloquium on Musical Informatics. Florence: AIMI,

pp. 45-50.

Mountain, T. 1980. "Sketches for the Analysis of Stria." Unpublished manuscript. John Chowning Papers, Stan-

ford University Archives.

Mountain, T. 1981. "A Notational Representation of John Chowning7s Stria." Unpublished manuscript. John Chowning Papers, Stanford University Archives. Mountain, T. 2005. Personal communication, April and August. Risset, J.-C. 1988. "Perception, environnement, musiques.77 Musique et Perception March: 10-42.

Risset, J.-C. 2001. "Problemes poses par Panalyse d7oeuvres musicales dont la realisation fait appel a l7informatique.

Analyse et creation musicales.77 Actes du Troisieme Congres Europeen d' Analyse Musicale de Montpellier. Paris: L7Harmattan/SFAM, pp. 131-160.

Risset, J.-C. 2005. "Sur l7impact de l7oeuvre scientifique, technique et musicale de John Chowning.77 In P. A. Cas- tanet et al., eds. John Chowning: Portraits polychromes.

Paris: INA/GRM and Michel de Maule, pp. 31-59. Riviere, Y., and C. Pouillon. 1976. Passage du XXe siecle.

Paris: IRCAM.

Roads, C. 1985. "John Chowning on Composition.77 In C. Roads, ed. Composers and the Computer. Los Altos,

California: William Kaufman, pp. 16-25. Also available on-line at .html (accessed 28 February 2007). Roads, C. 1996. The Computer Music Tutorial. Cam- bridge, Massachusetts: MIT Press.

Schoeller, P. 1986. "Mutation de Pecriture : Eclat, Stria,

Disintegrations Inharmoniques.77 In Inharmoniques. Le temps de mutations, n. 1. ed. Christian Bourgois.

Paris: IRCAM, pp. 197-208.

Zattra, L. 2003. "Science et technologie comme sources

d7inspiration au CSC de Padoue et a 17IRCAM de Paris.77

PhD thesis, Sorbonne-Paris IV University and Trento


$4 Computer Music Journal

This content downloaded from on Wed, 06 Apr 2016 05:28:59 UTC All use subject to